The ultimate goal of this research is to characterize the interactions directing RNA folding, and to use this knowledge to reliably predict structures of RNA molecules from their sequences. Since most diseases are mediated through RNA, and some viruses, including HIV, have RNA genomes, the results can lead to rational design of therapeutics. The results should also further the interpretation of sequences determined by the Human Genome Project and other sequencing efforts. The foundation for structure prediction will be provided by studies of the thermodynamic and structural properties of oligonucleotides. Particular emphasis will be placed on the sequence dependence of stability and structure for internal loops and multibranch loops. Secondary structure prediction by free energy minimization will be augmented with constraints from structure mapping and sequence comparison to partially compensate for incomplete knowledge of the factors affecting folding. The power of the method will be tested by attempting to determine the secondary structure of a messenger RNA.